Quite a racket!
In driving around with an HF mobile station in my vehicle I can hear these 200 kHz-spaced carrier groups almost everywhere around town during daylight hours - the roar getting much stronger in/near residential areas as you would expect. If driving through a residential neighborhood, it is very easy to tell when you drive past a house equipped with a SolarEdge PV system - and it is easily audible from a block or two away. Knowing the "fingerprint" of this PV system allows it to be identified uniquely - even at some distance.
Are they DX? 1
A question arose in my mind: Does this "grunge" produced by the SolarEdge PV systems propagate long distances?
To answer this question I checked a KiwiSDR at the Northern Utah WebSDR (link) - a site with which I am very familiar 2. This receive system is located about 3 miles (5km) from any residential area, bounded on three sides with mosquito-laden bird refuges (wetlands) and on the fourth side - the same as the closest houses - by a mountain. Additionally, the antenna used for the reception in Figures 3 and 4 below was the TCI-530 omnidirectional log-periodic (with circular polarization) - which does not have good gain at very low radiation angles, further precluding the reception of "nearby" PV systems via "ground wave".
The quick answer to the above question is YES - the roar of SolarEdge systems is propagated when conditions are "reasonable" 4 as shown in the screen capture below:
The signals represented by the "hump" in the highlighted portion of the analyzer plot in the top part of the image - and the "band" of noise on the waterfall display - between 14.199 and 14.200 MHz are the sum of the propagated low-level PV system carriers from... who knows where? To be clear, this energy is not likely to be from just one SolarEdge PV system and its individual optimizers (one for each panel) but more likely from the many thousands of such devices that are each, individually, radiating energy. What we are seeing is the total energy of the propagated systems, the frequency spread being centered around 14.1993 MHz.
It's worth noting that the fact that these signals do not
land on exactly the same frequency 5 - hence the Gaussian-like
distribution of energy - and this has interesting implications. Even though the
signal from each, individual optimizer is (more or less) a CW (unmodulated)
carrier, the fact that there are so many of them clustered together
means that, for statistical purposes, they might as well be a
distribution of noise energy: Unlike a with a single coherent CW
signal, the DSP filtering on modern radios will be able to do little/nothing to
reduce their effects if they were to cause interference due to its similarity to white noise.
A quick power and spectral analysis of the signal above showed that if the signals above were a single, coherent CW signal, the total amount of energy contained in the "hump" in Figure 3 would have easily been at least 15-20dB above the noise in a 50 Hz detection bandwidth: A CW signal of this strength would certainly be cause for complaints!
I also looked at other 200 kHz multiples around 14.000 and 14.400 and the same, exact types of signals were present on those frequencies - and similar bunches of energy fitting this profile were noted at least as low as 10.200 and as high as around 18.200 MHz as well (probably higher) and every (otherwise) clear frequency in between - this range being related to current ionospheric propagation at the moment that I checked (e.g. around 1845 on September 10 UTC, 2025). 6
To verify that these signals were propagated and were likely from SolarEdge systems, several things were done:
- The presence at many 200 kHz multiples/intervals across the HF spectrum is telling! Their being slightly below exact 200 kHz multiples as noted in Footnote 5 adds to their "uniqueness".
- On days with poor propagation overall, these signals were absent - or limited to frequencies commensurate with the MUF (Maximum Useable Frequency).
- These signals disappear at night. (This test is somewhat complicated by the fact that propagation on these bands also changes at night - but sunlight is still illuminating the ionosphere well after sunset on the ground.)
- An "S-meter" plot was run over the period of several minutes: A propagated signal(s) would show variations in signal strength - but this can be foiled to a degree by the fact that many, many individual point sources would each be propagated differently and unlike a single source, would not experience as deep a fading as the plot below shows:
As noted in the original article analyzing a system close-up (linked above) the SolarEdge optimizers produce other signals 6-10 dB weaker at various points above each 200 kHz interval - these are visible in Figure 1. When the above plots were made these signals weren't readily apparent - but I suspect that they will be visible during "excellent" propagation conditions rather than the "mediocre-to-average" conditions that were present when Figures 3 and 4 were produced.
Conclusion: They do get propagated!
So yes, you can DX SolarEdge PV systems - it's just that there are so many of them each doing their own radiating that you probably won't know from where those signals originate, so it's hard to know from how far away you might actually be hearing them! To be clear, it's difficult to determine if a the radiated RF from a single optimizer would be audible via ionospheric propagation, and with many thousands of them out there this may be impossible to determine - but it is clear that the summation of many thousands of them does produce an audible signal.
Do these signals actually cause QRM 7 ? As noted in the earlier post (liked above) they most certainly do if you live within a city block or two of one of the SolarEdge PV systems and operate on or near any of the frequencies occupied by the spurious radiation represented in Figure 1. If your receive system is located well away from a SolarEdge installation, the above shows that you may still experience interference from these systems - even from a significant distance.
Figure 3 also shows that the emissions do propagate over long distances: The 20 meter band's optimal "single skip" distance would likely place the majority of these signals in a 700-1500 mile (1100-2400 km) radius of Northern Utah - and this includes quite a few populated areas in parts of the U.S. where the number of solar installations is quite high.
You, too, can check for QRM at your station
If you have an HF station with a receiver with a waterfall display you might want to check the various amateur bands just below the 200 kHz multiples 8 during daylight hours: If there is a SolarEdge PV system within a couple city blocks of you 9 you will most likely see and hear it - but don't blame me if, after finding that you can see those signals, you can't "un-see" them!
* * *
Links to related pages (about solar power) on this blog:
- Analysis of a SolarEdge system (link) - This is the article linked at the top of the page where careful measurement was done to characterize the interference created by a SolarEdge system neighboring a local amateur.
- It *is* possible to have an RF quiet home PV system (link) - On this page I describe the SunnyBoy based system at my home QTH - and the fact that it does NOT produce QRM.
- Reducing QRM from a Renogy 200 watt (or any other) portable solar panel system (link) - Here I describe a portable solar system that I take "car camping" and how I reduce its interference to the point of inaudibility.
Footnotes:
- The term "DX" means distance. Generally speaking, if a signal is "DX" it is understood that it must be being propagated over much more than a line-of-sight distance - in this case, via ionospheric propagation at distances of hundreds or thousands of miles/km.
- The author of this post is one of the original founders and current maintainers of the Northern Utah WebSDR which has a remote HF receive site about 80 miles (94km) north of Salt Lake City.
- Figure 2 shows a "close-up" spectral view of the signals emitted by several SolarEdge PV systems within a mile/kilometer or two of my house - the closest system being about a block away. The center frequency of this cluster of signals was approximately 7.39965 MHz and a 256k-point FFT with a bin width of 183 mHz (milliHertz) - along with some averaging - was used to create this plot. Clearly visible are a large number of individual carriers along with a background "roar" of many more weaker carriers that are not individually distinguishable in this plot. This plot was purposely done on a frequency above the 40 meter amateur band during daylight hours (the local time is visible in the image) and during this time there is no strong, long distance propagation (a fact verified by the absence of a similar set of signals on the remote Northern Utah WebSDR site) indicating that this energy is, in fact, originating from systems proximate to my own receive site. At sunset, these carriers will gradually disappear - often "blinking" out - as the solar panels lose their light and will reappear the next morning: This "blinking" can be heard as individual tones flicker on/off during the day<>night transition by listening on an ordinary SSB-capable receiver at one of the frequencies noted above.
- The frequencies mentioned have also been checked when ionospheric propagation is poor (comparatively few strong signals) and the characteristic SolarEdge carriers were absent at the remote receive site. This further illustrates the fact that the signals described above are not local to the remote receive site and reinforces the likelihood that they are, in fact, being propagated.
- Observation of a SolarEdge PV system at very close distance (less than 50 feet/15 meters) indicates that each, individual optimizer - a device attached to the back of every individual solar panel - will radiate the signals at 200 kHz intervals. Due to the slight variations in oscillator frequencies (e.g. quartz crystals or MEMs devices) the precise frequencies of these signals - and their harmonics - will vary, but the mean frequency separation appears to be around 199.9901 kHz which puts them slightly below a precise 200 kHz multiple which is why the peak of the distribution shows up around 14.1993 MHz on 20 meters, 7.19965 MHz on 40 meters and so on. As noted in the text, the actual frequency spread of the individual modules is such that it has a Gaussian-like distribution above and below the mean frequency.
- I also checked several remote receive systems around the world during their local daylight hours and could see the same "humps" of energy at frequencies just below the aforementioned 200 kHz multiples on some of them. One such system was that located at the University of Twente in the Netherlands: It is not known to what degree the signals that were radiated (likely) from PV systems were propagated and which might be within a few kilometers of this receive site, but they are certainly "there".
- "QRM" is a "Q" signal referring to "Man Made Interference" and the magnitude of this interference in comparison to the desired signals determines if this is harmful interference. If QRM makes it difficult/impossible to receive a signal on frequency, that would fit the definition of harmful interference.
- The frequencies on which the radiated signals from a SolarEdge PV system (every 199.9901 kHz) will likely land within an HF amateur band are clustered around the following: 3.5998, 3.7998, 3.9998, 7.1996, 14.1993, 21.1990, 21.3990, 28.1986, 28.3986, 28.5986, 28.7986, 28.9986, 29.1986, 29.3986 and 29.5986 MHz plus similar frequencies in the 6 meter band: They can also be heard on non-amateur frequencies at the same 199.9901 kHz intervals as well. As the above frequencies are the actual frequencies, you will need to tune above or below the frequencies (using LSB or USB, respectively) by 1.5 kHz or so to hear the "roar". Of course, you will only hear these signals during daylight hours when the PV systems are active. Note that the combination of naturally-higher noise levels on the lower bands (80, 40 meters) and the likely lower efficiency of the PV system's component ability to radiate RF there - plus the tendency for nighttime propagation on those bands (when the PV systems are inactive) - means that observing this phenomenon on those frequencies via the ionosphere is much less likely.
- If you do some remote operation like POTA or SOTA at a significant distance from any likely PV system, you might want to take a look at one of the 200kHz-interval frequencies mentioned above during daylight hours and good propagation: You'll probably see the propagated PV signals there, too.
* * * * *
This page stolen from ka7oei.blogspot.com
[END]